A SKUs processing algorithm
Today I’m going to tell you how we created for one of our clients, a large hardware manufacturer in Europe, a SKUs processing algorithm, which allowed us to sort orders by the order tracking system. Owners of industrial enterprises and those involved in the issues of work processes automation, in principle, will find it interesting.
Input data
Client: Well-known manufacturer of fasteners in Italy, shipping its products to more than 30 countries. Wholesale supplies, tens of thousands of contractors, high average bill.
The problem is the complexity of order processing, the large amount of time for processing, and insufficient understanding of how to correctly enter CRM data. An example is a case where managers failed to provide fully prepared hot leads to sale: the audit discovered more than 700 applications that did not get to the right department but were lost along the way. Lesser global failures were also found when managers accepted an order for a product that was out of stock, or could not find available products; and the isolation of telephone and electronic communications of each individual company office in different cities and countries hindered their coordinated work.
Task: to completely modify the algorithm of work processes, making them less time-consuming and effort-consuming for employees and be more efficient.
Situation analysis and work plan
We understood that to solve the problem it was necessary to make adjustments to the very structure of the company’s activities, as well as to develop and successfully implement a CRM system that would automate the main production processes and take outsourced customer success solutions to the new level.
We identified the main KPIs that we should strive for and outlined several main steps:
- Ensure that the client receives a commercial offer within 2 hours from the time of placing the order.
- Implement a new CRM with digitized request processing.
- Automate the activities of operators loading orders into 1C.
- Combine CRM with the existing 1C database and matching service.
One of the weaknesses of the old system was the inability for employees to manually collate orders received daily from a huge number of counterparties in different formats effectively. Due to the lack of a successful automated operator interface, the highest level of outsource customer service could be forgotten. The process looked something like this:
Everything happened clumsily, the sequence of actions was constantly disrupted, and the order files, which the operators manually collected, correlated with the internal nomenclature and prepared by entering into 1C, were regularly confused and lost. Adding even more chaos was the fact that the client could call the same item differently, adding to the misunderstanding among managers:
Clearing confusion
Our analysts identified two more features in the company's work that hindered its development. Firstly, CRM, widely used for consulting companies, here served only as a client base, and the system was not used to conduct transactions. Secondly, counterparties transmitted order data to operators in a wide variety of formats, including voice messages and photos. It was necessary to reorganize sales as a service, convenient for customers and profitable for the company.
We decided to start by completely breaking the logic of the old ordering system, reformatting it as follows:
After the changes we made, order processing began to occur through the item matching service, while simultaneously checking the product balances in the warehouse and transferring data to CRM for transaction analysis. Now these processes have acquired a more orderly form:
You don’t need to have a deep knowledge of marketing, much less be a bpo specialist, to see the advantages of the new scheme over the old one, where everything often happened on a random basis.
Choosing a CRM platform
To successfully implement CRM, it was first necessary to develop a system that would meet all the needs of the customer’s company. At first, we planned to use Corteza CRM to build its processes, but in the end, we decided that the Terrasoft platform was much more suitable in this case, since all the scripts and widgets necessary were available for use to carry out complex business processes.
Why is it important?
Let's take an example of the really large, best call centers, where up to five thousand operators work. Customer calls in such centers are segmented into several types, each of which has its own conversation scenario. A cheap CRM will not be able to cope with such a load and will simply confuse managers, while a clear algorithm provides the necessary flexibility in scenarios and branching of processes will show itself in the best way.
Now all customer orders, including those who contact the company for the first time, enter a single system, where managers receive them and lead them through to placing the order, and the process is automatically transferred through all the necessary stages and transferred to the necessary employees for execution. As a result, the transaction is under control at any time, and it becomes easier for operators and managers to perform their duties.
If you have a small company, you can turn to a simpler platform to implement CRM, but if your business does not fit into the small framework, or you are planning to expand, I recommend betting on something more serious, capable of handling large-scale workloads. For example, Terrasoft, which is used by employees in my agency and, as I know, by many US companies that outsource.
Updated matching system
Once the CRM was ready, it was time to move on to the next task. “Manual” order matching took too much time for operators – sometimes up to several days, so it was clear that this process also needed to be automated. Here, we resorted to Elasticsearch, a search engine that converts source text into unique identifiers and tokens, and then compares them with the existing database and ultimately outputs the finished product positions.
Using tokenization as a basis, we created the Elastic system, which performs the work of operators in matching orders and their outsourced sales ready, after which it uploads the finished result to 1C, and the human task is only to control the actions of Elastic and its further training.
Two types of lists
One of the training points is the compilation of black and white lists:
- If the chain Client – Item from document – Nomenclature from 1C has been matched, it is whitelisted and reused;
- If it turns out to be incorrect, it will be blacklisted and not be used further.
Over time, the list will expand, ensuring increasingly accurate operation of the system in automatic mode. You get a kind of sales virtual assistant, although, of course, this is not exactly what this term is usually understood as.
After matching the list of positions of the received file, its data is displayed in the form of a table:
All that is required of the operator is to make sure that all records are matched correctly in order to train the system.
Underwater rocks
Naturally, it was not without difficulties. And the first of them during the assembly of Elastic was the enterprise’s product catalog that had not been updated for a long time. When you realize that it contains 4 times more items than are actually in the company’s warehouse, this is a problem that needs to be solved as soon as possible, otherwise all work will stop. Therefore, I had to quickly do an update. Next, we introduced a unified packaging and volume system for orders to eliminate possible discrepancies. Everything became simple and clear.
From there, the system converted files that clients each prepared in their own format into one common one, recognizing and isolating fundamental parameters from it.
The last rather unexpected problem was the resistance from operators, who began to fear that they would be fired. A small spoiler: after the final implementation of the system, the management plans to transfer some of them to support, and some to the ranks of account managers, so those concerns are for naught.
What did we end up with?
Development and implementation of CRM with all innovations took our team about 3 months. At the moment, we have a working MVP, the system is being successfully tested in one of the company’s divisions, after which it will be extended to all others.
The digitization of each stage of the order is being completed, from the customer’s request to the delivery of the goods, which has already made it possible to relieve managers and give them the opportunity to pay more attention and effort to issues of best customer service. Indicators of their activity are now also sent to CRM, clearly demonstrating cross-sections of the work of each employee.
Work continues to train and improve Elastic.
Work is still ongoing to improve the quality of delivery customer service; the ideal level of process and cost optimization indicators has not been achieved. All this requires additional effort, but it will certainly pay off. According to forecasts, by the time of full implementation, the system will save the enterprise up to 450 thousand per month.
I am sure that it is possible to automate any business process – from mailing to accounting outsource if you have a good idea of the benefits it will bring. Think about it: do you need to improve processes that are already running? Is it worth building a fundamentally new business model from scratch? Will optimization pay off? Answer these questions for yourself and, if you understand that yes, act.